Updating vegetation classifications: an example with New Zealand's woody vegetation
Article first published online: 5 JUL 2012
© 2012 International Association for Vegetation Science
Journal of Vegetation Science
Volume 24, Issue 1, pages 80–93, January 2013
How to Cite
Wiser, S. K., De Cáceres, M. (2013), Updating vegetation classifications: an example with New Zealand's woody vegetation. Journal of Vegetation Science, 24: 80–93. doi: 10.1111/j.1654-1103.2012.01450.x
- Issue published online: 4 DEC 2012
- Article first published online: 5 JUL 2012
- Manuscript Accepted: 29 MAY 2012
- Manuscript Received: 28 SEP 2011
- New Zealand Department of Conservation
- New Zealand Ministry of Science and Innovation. Grant Number: C09X0916
- Beatriu de Pinós. Grant Number: BP-B 00342
- Catalan Agency for Management of University and Research
- Consolider Montes. Grant Number: CSD2008-00040
- Spanish Ministry of Education and Science (MEC)
- Royal Society of New Zealand. Grant Number: SPN 10-13
- Community ecology;
- Fuzzy classification;
- Noise clustering;
- National Vegetation Survey (NVS) databank;
- Vegetation databases
How can existing vegetation classifications be updated when new plot data are obtained? Can we use the properties of plots classed as outliers to identify gaps in our understanding of vegetation patterns and so direct future enquiry?
We updated a pre-existing classification of New Zealand's forests and shrublands based on a nationally representative data set (1177 plots) by using 12 374 additional plot records from New Zealand's National Vegetation Survey Databank (NVS). We resampled the NVS plot records to remove uneven representation along floristic and geographic gradients. To update the classification at the alliance level, we first cast the original classification into the fuzzy classification framework of Noise Clustering and then discarded original types with low plot numbers and high compositional variation. We then used the plot records that could not be assigned to any original alliance to define new alliances, while retaining the original alliances as fixed elements. We also defined vegetation associations to create a classification at a lower level of abstraction and related it to the classification at the alliance level. Finally, we determined whether known rare types were represented among the new vegetation types and characterized plot records classed as outliers.
After casting the 24 original alliances in the NC framework, we discarded seven. We extended the 17 remaining alliances with 12 new ones and defined 79 associations. All 12 new alliances had extents <120 986 ha, which is smaller than the original alliances, and included rare types that were known to exist but could not be defined using the objectively sampled data set underpinning the original classification. Plot records classed as outliers tended to occur at lower altitudes or in successional shrublands. Further sampling is required to adequately define vegetation types in such situations, although composition may be inherently erratic in successional shrublands.
Our analysis illustrates the application of a fuzzy classification framework at a national scale and provides a model for others wishing to extend and update vegetation classifications. Our approach allows rare community types to be defined and identifies portions of compositional and geographic gradients that are poorly documented.